O M B N O . 0 7 0 4 -0 1 8 8 P u b lic reporting burden fo r this colle ction o f in form atio n is e stim a te d to a verag e 1 h ou r per response, in clu din g th e tim e fo r re vie w in g in stru ction s s e a r c h i n g^^ o a th e rin q a n d m aintaining the data needed, and co m p letin g a nd re vie w in g th e co lle ctio n o f inform ation. S end com m e nt re g ard in g th is burde n e stim a te o r a ny o th e r a spect o f this collection o f inform ation, including su g g estio n s fo r re d u d n g this burden, to W a sh in g to n H ead q ua rte rs S ervices. D n c M f°r n a v is H in h w a v S uite 1 2 0 4 . A rlin a to n . V A 2 2 2 0 2 -4 3 0 2 , and to th e O ffice o f M an ag em en t and B udget, P ap e rw o rk R edu ctio n P ro ject (0 7 0 4 -0 1 8 8 ), W a sh ing ton , D O 2 0 0 0 3 .2. REPORT DATE 3. REPORT TYPE AND DATES COVERED S ep te m b er, 2 0 1 0
TITLE AND SUBTITLET o w ard s C o v erag e C lo su re: U sin g G o ld M in e A sse rtio n s fo r G e n e ra tin g D esig n V alid atio n S tim ulus W e p resen t a m e th o d o lo g y to g e n e ra te in p u t stim u lu s fo r d esig n v a lid a tio n u sin g G o ld M in e , an au to m a tic asse rtio n g en e ratio n en g in e th a t uses data m in in g an d fo rm a l v e rific a tio n . G o ld M in e m in e s th e sim u latio n tra c e s o f a b e h a v io ra l R e g iste r T ra n sfe r L ev el (R T L ) d e s ig n u sin g a d ecisio n tre e b a s e d le a rn in g a lg o rith m to p ro d u c e c a n d id a te assertio n s. T h e se ca n d id a te a sse rtio n s are p assed to a fo rm a l v erificatio n eng in e. I f a c a n d id a te asse rtio n is false, a c o u n te re x a m p le tra c e is g e n e ra te d . In th is w o rk , w e fe e d these co u n terex am p le trac es to ite ra tiv e ly re fin e th e o rig in a l sim u la tio n tra c e data. W e in tro d u c e an in c re m e n ta l d e c isio n tree to m in e the n e w traces in each iteratio n . T h e alg o rith m c o n v e rg e s w h e n all th e c a n d id a te a sse rtio n s are tru e. W e p ro v e th a t o u r alg o rith m w ill alw a y s co n v erg e an d ca p tu re th e c o m p le te fu n c tio n a lity o f an o u tp u t o n co n v e rg en ce . W e sh o w th a t o u r m e th o d alw a y s resu lts in a m o n o to n ic in crease in sim u la tio n co v e rag e. W e also p re se n t an o u tp u tc en tric n o tio n o f c o v e ra g e , an d a rg u e th a t w e ca n attain co v e rag e clo su re w ith re sp e c t to th is n o tio n o f co v e rag e. E x p e rim e n ta l re su ltsto v a lid a te o u r a rg u m e n ts are p re se n te d on R ig el, a 1 0 0 0 + core p ro ce sso r d esig n as w ell as se v e ra l o th e r b e n c h m a rk s.14. SUBJECT TERMS D a ta m ining, v erificatio n , te s t g en e ratio n , C o v e ra g e {Iiu187, dsherid3, tuohy2, shobhav}@illinois.edu
ABSTRACTWe present a methodology to generate input stimulus for design validation using GoldMine, an automatic assertion generation en gine that uses data mining and formal verification. GoldMine mines the simu...